Constructions of generalized bent–negabent functions Aditi Kar Gangopadhyay and Ankita Chaturvedi Department of Mathematics, Indian Institute of Technology Roorkee Roorkee 247667 INDIA, {ganguli.aditi, ankitac17}@gmail.com
Abstract. In this paper, we consider generalized Boolean functions from Zn 2 to Zq . We describe two constructions of generalized bent–negabent functions.
Keywords: Generalized Boolean functions, nega–Hadamard transform, bent– negabent functions.
1
Introduction
Let Z, R and C denote the set of integers, real numbers and complex numbers respectively. Any element x ∈ Zn2 can be written as an n-tuple (xn , . . . , x1 ), where xi ∈ Z2 for all i = 1, . . . , n. The addition over Z, R and C is denoted by ‘+’. The addition over Zn2 for all n ≥ 1, is denoted by ⊕. Addition modulo q is denoted by ‘+’ and is understood from the context. If x = (xn , . . . , x1 ) and y = (yn , . . . , y1 ) are two elements of Zn2 , we define the scalar (or inner) product, respectively, the intersection by x · y = xn yn ⊕ · · · ⊕ x2 y2 ⊕ x1 y1 , x ∗ y = (xn yn , . . . , x2 y2 , x1 y1 ). The√ cardinality of the set S is denoted by |S|. If z = a + b ı ∈ C, then |z| = a2 + b2 denotes the absolute value of z, and z = a − b ı denotes the complex conjugate of z, where ı2 = −1, and a, b ∈ R. The conjugate of a bit b will also be denoted by ¯b. A function from Zn2 to Z2 is said to be a Boolean function on n variables and the set of all such functions is denoted by Bn . For more detail theory of Boolean functions we refer [1–3]. A function from Zn2 to Zq (q, a positive integer) is said to be a generalized Boolean function on n variables [10] and we denote the set of all such functions by Bnq . The (generalized) Walsh–Hadamard transform of f ∈ Bnq at any point u ∈ Zn2 is the complex valued function defined by X n Hf (u) = 2− 2 ζ f (x) (−1)u·x . x∈Zn 2
2πı
where ζ = e q is the q-th primitive root of unity. A function f ∈ Bnq is said to be generalized bent if and only if |Hf (u)| = 1 for all u ∈ Zn2 . The nega–Hadamard transform of f ∈ Bn at any vector u ∈ Zn2 is the complex valued function X n Nf (u) = 2− 2 (−1)f (x)⊕u·x ıwt(x) . x∈Zn 2
A function f ∈ Bn is said to be negabent if and only if |Nf (u)| = 1 for all u ∈ Zn2 . For further study of nega-Hadamard transforms and negabent functions we refer to [5–7, 11, 8, 13]. Definition 1. The nega–Hadamard transform of f ∈ Bnq at any point u ∈ Zn2 is defined by X n Nfq (u) = 2− 2 ζ f (x) (−1)u·x ıwt(x) . x∈Zn 2
Definition 2. A function f ∈ Bnq is a generalized negabent function if |Nfq (u)| = 1 for all u ∈ Zn2 . We shall use throughout the well-known identity (see [4]) wt(x ⊕ y) = wt(x) + wt(y) − 2wt(x ∗ y).
2
(1)
Existence of a subclass of generalized bent–negabent functions in generalized Maiorana–McFarland class
Schmidt [12] provided the generalization of Maiorana–McFarland construction of Z2 -valued bent functions to Z4 -valued bent function. The generalization is given in Lemma 1. St˘ anic˘ a et al. [9] generalized it for Zq -valued (for any even q) bent function. They refereed it by generalized Maiorana–McFarland class (GMMF). The GMMF is given in Lemma 2. Lemma 1. [12, Thm. 5.3] Let g : Zn2 → Z4 be arbitrary. Any function f : Z2n 2 → Z4 be given by f (x, y) = 2σ(x) · y + g(x), for all x, y ∈ Zn2 where σ is a permutation on Zn2 , then f is generalized bent Boolean function. Lemma 2. [9, Thm. 9] Suppose q is an even positive integer. Let σ be a permutation on Zn2 , let g : Zn2 → Zq be an arbitrary function, then the function f : Z2n 2 → Zq defined as q y · σ(x) for all x, y ∈ Zn2 (2) f (x, y) = g(x) + 2 is a generalized bent Boolean function.
Let σ is weight-sum invariant permutation on Zn2 , i.e., it satisfies by the following relation, wt(x ⊕ y) = wt(σ(x) ⊕ σ(y)) for all x, y ∈ Zn2
(3)
Noticed that if σ is orthogonal, i.e., σ(x) = A·x where A is orthogonal (i.e., AT = A−1 ), then it satisfies the imposed condition (since wt(σ(x) ⊕ σ(y)) = wt(A(x ⊕ y)), it suffices to show that wt(Az) = wt(z); for that, consider wt(Az) = (Az)T · (Az) = zT (AT A)z = wt(z)). Theorem 1. Any function, f (x, y) = 2q y · σ(x) + g(x), for all x, y ∈ Zn2 , as in (2) on Z2n 2 is generalized bent–negabent if and only if g is generalized bent. Proof. We evaluate q
ζ g(x)+( 2 )y·σ(x) (−1)x·u⊕y·v ıwt(x)+wt(y)
X
Nfq (u, v) = 2−n
(x,y)∈Z2n 2
X
= 2−n
x∈Zn 2
X
= 2−n
X
ζ g(x) ıwt(x) (−1)x·u
(−1)(σ(x)⊕v)·y ıwt(y)
y∈Zn 2
ζ g(x) ıwt(x) (−1)x·u 2n/2 ω n ı−wt(σ(x)+v)
x∈Zn 2
= 2−n/2 ω n
X
ζ g(x) (−1)x·u ıwt(x)−wt(σ(x)+v) .
x∈Zn 2
Now, using the fact that σ is a weight-sum invariant permutation, and by (1), we obtain wt(σ(x) ⊕ v) = wt(x ⊕ σ −1 (v)), wt(x) − wt(σ(x) ⊕ v) = −wt(σ −1 (v)) + 2wt(x ∗ σ −1 (v)), and ı2wt(x∗σ
−1
(v))
= (−1)x·σ
−1
(v)
,
which implies that Nfq (u, v) = 2−n/2 ω n ı−wt(σ
−1
(v))
X
ζ g(x) (−1)(u⊕σ
−1
(v))·x
x∈Zn 2
= ω n ı−wt(σ
−1
(v))
Hg (u ⊕ σ −1 (v)).
Consequently, |Nfq (u, v)| = |Hg (u ⊕ σ −1 (v))|, which implies our claim. t u
3
Construction of generalized bent–negabent functions
Schmidt et al. [13] described construction of bent–negabent functions. In this section, we provide the characterization of generalized bent–negabent functions for arbitrary positive even integer q. Let us define V be mn dimensional vector space over Z2 , so that, V = Vn ⊕ Vn ⊕ . . . ⊕ Vn . | {z } m−times
Let the generalized Boolean function f : V ⊕ V → Zq expressed as follows f (x1 , . . . , xm , y1 , . . . , ym ) q σ(x1 , . . . , xm ) · (y1 , . . . , ym ) = g(x1 , . . . , xm ) + 2 where q is positive even integer, the function σ : V → V is of the form
(4)
σ(x1 , . . . , xm ) = (ψ1 (x1 ), φ1 (x1 ) ⊕ ψ2 (x2 ), . . . , φm−1 (xm−1 ) ⊕ ψm (xm )) and g : V → Zq is defined by g(x1 , . . . , xm ) = h1 (x1 ) + h2 (x2 ) + . . . + hm (xm ). Here, ψ1 , . . . , ψm , φ1 , φm−1 are permutations on Vn and h1 , . . . , hm : Vn → Zq be arbitrary generalized Boolean functions, i.e., f reads as q f (x1 , . . . , xm , y1 , . . . , ym ) = h1 (x1 ) + ψ1 (x1 ) · y1 2 m q X (φj−1 (xj−1 ) ⊕ ψj (xj )) · yj + hj (xj ) + 2 j=2 Since σ is a permutation. Therefore f is Generalized Maiorana–McFarland (GMMF) type generalized bent function. t u Lemma 3. [13, Equation 3, 4] Let m be a positive integer satisfying m 6≡ 1(mod 3), and let k be an integer satisfying 0 < k < m and k ≡ 0(mod 3) or (m − k) ≡ 1(mod 3). Then P (z2k ) = 2kn ω c δψ(z2k )⊕a ,
for some c ∈ Z8 , a ∈ Vn
and Q(z2k ) =
2(m−k)n ω d (−1)b·φ(z2k ) , if m ≡ 0 (mod 3), (m−k)n d b·φ(z2k ) −wt(ψ(z2k )) 2 ω (−1) ı if m ≡ 2 (mod 3).
for some d ∈ Z8 , b ∈ Vn , where φ, ψ are permutations on Vn and ω = 8-th primitive root of 1.
1+ı √ 2
is an
P (z2k ) = 2k−2 Y
X
j=1 zj ∈Vn
(−1)(zj+1 +uj )·zj ıwt(zj )
X z2k−1 ∈Vn
(−1)(ψ(z2k )+u2k−1 )·z2k−1 ıwt(z2k−1 )
and Q(z2k ) = 2m Y
X
(−1)(zj−1 +uj )·zj ıwt(zj )
j=2k+2 zj ∈Vn
X
(−1)(φ(z2k )+u2k+1 )·z2k+1 ıwt(z2k+1 ) .
z2k+1 ∈Vn
t u
Theorem 2. Let m be a positive integer satisfying m 6≡ 1(mod 3), and let k be an integer satisfying 0 < k < m and k ≡ 0(mod 3) or (m − k) ≡ 1(mod 3). Let f be defined as in (4), where σ(x1 , . . . , xm ) = (x1 , x1 + x2 , xk−1 + ψ(xk ), φ(xk ) + xk+1 , . . . , xm−1 + xm ) and g(x1 , . . . , xm ) = h(xk ). φ, ψ are permutations on Vn and h : Vn → Zq be arbitrary generalized Boolean function. Then f is generalized bent–negabent. Proof. Define by relabeling z2j := xj and z2j−1 := yj for 1 ≤ j ≤ m, so that we have f (x1 , . . . , xm , y1 , . . . , ym ) = 2k−2 2m X q X zj · zj+1 + z2k−1 · ψ(z2k ) + zj · zj−1 + z2k+1 · φ(z2k ) . h(z2k ) + 2 j=1 j=2k+2
2πı q
Let ζ = e
. Now, compute
Nfq (u1 , . . . , u2m ) = 2−mn
X
ζ f (z1 ,...,z2m ) (−1)(z1 ,...,z2m )·(u1 ,...,u2m ) ıwt(z1 ,...,z2m )
z1 ,...z2m ∈Vn −mn
=2
X
ζ h(z2k ) (−1)z2k ·u2k ıwt(z2k ) P (z2k )Q(z2k ).
z2k ∈Vn
Using Lemma 3 we have Nfq (u1 , . . . , u2m ) c+d P h(z2k )+ q2 (z2k ·u2k +b·φ(z2k )) wt(z2k ) ı δψ(z2k )⊕a) , if m ≡ 0 (mod 3), ω z2k ∈Vn ζ P = c+d h(z2k )+ q2 (z2k ·u2k +b·φ(z2k )) ω δψ(z2k )⊕a if m ≡ 2 (mod 3). z2k ∈Vn ζ ( q −1 −1 −1 −1 h(ψ (a))+ (ψ (a))·u +b·φ(ψ (a)) ) ıwt(ψ (a)) , if m ≡ 0 (mod 3), 2k 2( ω c+d ζ = q −1 −1 −1 c+d h(ψ (a))+ 2 ((ψ (a))·u2k +b·φ(ψ (a))) if m ≡ 2 (mod 3). ω ζ (5) Therefore, |Nfq (u1 , . . . , u2m )| = 1 for all (u1 , . . . , u2m ) ∈ V2m n . Hence we conclude our result. t u Example 1. Take m = 2 and k = 1 in Theorem 2. Then f becomes q f (x1 , x2 , y1 , y2 ) = h(x1 ) + {y1 · ψ(x1 ) + φ(x1 ) · y2 + y2 · x2 }. 2 Thus we can construct generalized bent–negabent functions in 4n variables of degree ranging from 2 to n.
4
Conclusion
In this paper, we provide a construction on generalized bent–negabent functions. Moreover, we identify a class of generalized bent–negabent functions.
References 1. C. Carlet, Boolean functions for cryptography and error correcting codes. In: Y. Crama, P. Hammer (eds.), Boolean Methods and Models, Cambridge Univ. Press, Cambridge. Available: http://www-roc.inria.fr/secret/Claude.Carlet/pubs.html. 2. T. W. Cusick, P. St˘ anic˘ a, Cryptographic Boolean functions and Applications, Elsevier–Academic Press, 2009. 3. R. Lidl, H. Niederreiter, Introduction to finite fields and their applications, Cambridge University Press, 1983. 4. F. J. MacWilliams, N. J. A. Sloane, The theory of error–correcting codes, NorthHolland, Amsterdam, 1977. 5. M. G. Parker, A. Pott, On Boolean functions which are bent and negabent. In: S.W. Golomb, G. Gong, T. Helleseth, H.-Y. Song (eds.), SSC 2007, LNCS 4893 (2007), Springer, Heidelberg, 9–23. 6. C. Riera, M. G. Parker, One and two-variable interlace polynomials: A spectral interpretation, Proc. of WCC 2005, LNCS 3969 (2006), Springer, Heidelberg, 397– 411. 7. C. Riera, M. G. Parker, Generalized bent criteria for Boolean functions, IEEE Trans. Inform. Theory 52:9 (2006), 4142–4159. 8. P. St˘ anic˘ a, S. Gangopadhyay, A. Chaturvedi, A. K. Gangopadhyay, S. Maitra, NegaHadamard transform, bent and negabent functions, in Pro. of SETA 2010, LNCS 6338 (2010), pp. 359–372. 9. P. Stanica, S. Gangopadhyay, B. K. Singh Some Results Concerning Generalized Bent Functions, http://eprint.iacr.org/2011/290.pdf. 10. P. Sol´e, N. Tokareva, Connections between Quaternary and Binary Bent Functions, http://eprint.iacr.org/2009/544.pdf. 11. S. Sarkar, On the symmetric negabent Boolean functions, Indocrypt 2009, LNCS 5922 (2009), 136–143. 12. K-U. Schmidt, Quaternary Constant-Amplitude Codes for Multicode CDMA, IEEE International Symposium on Information Theory, ISIT’2007 (Nice, France, June 24–29, 2007), 2781–2785; available at http://arxiv.org/abs/cs.IT/0611162. 13. K. U. Schmidt, M. G. Parker, A. Pott, Negabent functions in the Maiorana– McFarland class. In: S.W. Golomb, M.G. Parker, A. Pott, A. Winterhof (eds.), SETA 2008, LNCS 5203 (2008), Springer, Heidelberg, 390–402.